IJ50516: HA/GLVM: TAKEOVER AFTER SITE FAILURE FAILS IF NEW LV WAS ADDED

APAR status

• Closed as program error.

Error description

• A takeover event fails in a PowerHA/GLVM cluster
  (non-Enhanced Concurrent Mode VGs are used) after
  a site failure if a new LV has been created before.
  
  Note:
  In 2-node cluster a node failure is a site failure.
  
  hacmp.out of remaining/takeover node looks like:
  ...
  +clCglvmrg:clvaryonvg(24.637):asyncGLVMvg1[1328]
   [[ 1667923571843708220 != 636a7e733249f73c ]]
  +clCglvmrg:clvaryonvg(24.637):asyncGLVMvg1[1331]
   : The timestamps on at least one readable disk
   does not match
  +clCglvmrg:clvaryonvg(24.637):asyncGLVMvg1[1332]
   : that contained in the ODM. Tell LVM to update
   its local
  +clCglvmrg:clvaryonvg(24.637):asyncGLVMvg1[1333]
   : information from the disk
  +clCglvmrg:clvaryonvg(24.637):asyncGLVMvg1[1335]
   [[ -z hdisk1 ]]
  +clCglvmrg:clvaryonvg(24.637):asyncGLVMvg1[1339]
   importvg -L asyncGLVMvg1 -R hdisk1
  0516-1287 varyonvg: IOCINFO ioctl for /dev/hdisk8 failed.
  0516-780 importvg: Unable to import volume group from
   hdisk1.
  ...
  +clCglvmrg:clvaryonvg(26.921):asyncGLVMvg1[1449]
   varyonvg -n -t -O asyncGLVMvg1
  ...
  +clCglvmrg:clvaryonvg(27.262):asyncGLVMvg1[1449]
   varyonvg_output=$'+clCglvmrg:clvaryonvg(26.922):
   asyncGLVMvg1[1449]
   LC_ALL=C\n0516-052 varyonvg: Volume group cannot
    be varied on without a\n
   \tquorum. More physical volumes in the  group must
     be active.\n
   \tRun diagnostics on inactive PVs.'
  +clCglvmrg:clvaryonvg(27.262):asyncGLVMvg1[1450]
   varyonvg_rc=20
  ...
  +clCglvmrg:cl_mirrorset(1.964):asyncGLVMvg1[299]
   lslv -L -m -n hdisk1 asyncGLVMvg1t4
  0516-306 lslv: Unable to find  asyncGLVMvg1t4
   in the Device
          Configuration Database.
  ...
  +clCglvmrg:cl_mirrorset(2.101):asyncGLVMvg1[394]
   : All attempts to read partition map of LV
     asyncGLVMvg1t4 failed.
  ...
  +clCglvmrg:cl_mirrorset(2.101):asyncGLVMvg1[443] return 1
  +clCglvmrg:clvaryonvg(29.388):asyncGLVMvg1[1544]
   : Force is not an option, or has already been tried.
  +clCglvmrg:clvaryonvg(29.388):asyncGLVMvg1[1546] exit 20
  +clCglvmrg:cl_activate_vgs(29.689):asyncGLVMvg1
   [vgs_chk:104] RC=20
  +clCglvmrg:cl_activate_vgs(29.689):asyncGLVMvg1
   [vgs_chk:107]
   (( 20 == 1 || 20 == 20 ))
  +clCglvmrg:cl_activate_vgs(29.689):asyncGLVMvg1
   [vgs_chk:111]
   cl_RMupdate resource_error asyncGLVMvg1 cl_activate_vgs
  2022-11-08T17:11:27.849767
  ...
  

Local fix

Problem summary

• A takeover event fails in a PowerHA/GLVM cluster (non-Enhanced
  Concurrent Mode VGs are used) after a site failure. This happens
   because of an LV is created before site failure and lazy update
   is not triggered during failover.
  

Problem conclusion

• Code changes are done to perform an exportvg and a forced
  importvg so that a lazy update is triggered for new LV during
  failover operation.
  

Temporary fix

Comments

APAR Information

• APAR is sysrouted FROM one or more of the following:

  IJ44569

• APAR is sysrouted TO one or more of the following:

  IJ51075

Fix information

• Fixed component name

  POWERHA SYSMIR

• Fixed component ID

  5765H3900

Applicable component levels